Suppression of noise in optical associative memories by real time thresholding
Additional Document Info
Associative memories that recognize a pattern based on partial input have numerous applications such as homeland security. Optical implementations of associative memories, for example using computer-generated holograms, have the inherent parallelism as an advantage over software realizations. The nonlinear thresholding operation is a key step in the optical associative memories. A major source error in these memories is the thresholding uncertainties caused by fluctuation, for example in the input illumination or varying degrees of partial obscuration. Here, we show a proof-of-principle demonstration of a new scheme to suppress such errors using real time thresholding and a modified Hopfield associative memory model.
name of conference
Fluctuations and Noise in Photonics and Quantum Optics III